siRSM (version 1.1)

siRSM-package: siRSM: Single-index Response Surface Models

Description

This package provides tools for fitting what we call "single-index response surface models", that is, models of the form y=f(u,v)+e, where f(u,v) ~ 1 + u + v + u^2 + uv + v^2, u=t(w)%*%U, and v=t(w)%*%V.

Details

Package:

siRSM

Type:

Package

Version:

1.1

Date:

2014-07-15

License:

GPL-2

References

Cheng H, Zhu M, Chan VW, Michela, JL (in preparation), "Single-index response surface models".

See Also

siRSM, surface.test, surface.stats, ci.index, ci.surface

Examples

## Not run: 
# # load illustrative data set provided within the package
# data(EdwardsMBA)
# 
# # parse the variables
# y=EdwardsMBA[,'AVGSAT4']
# U=EdwardsMBA[,c('PCPRE','DMPRE','EIPRE','MRPRE')]
# V=EdwardsMBA[,c('PCACT','DMACT','EIACT','MRACT')]
# 
# # fit the model
# m1=siRSM(y,U,V)
# 
# # look at the model
# m1
# plot(m1)
# 
# # F-test of surface curvature
# surface.test(m1)
# 
# # inference on the index (using just 10 bootstrap samples here)
# ci.index(y,U,V,B=10)
# 
# # obtain statistics of the response surface (conditional on the estimated index)
# surface.stats(m1)
# 
# # inference for these surface statistics (again, conditional on the estimated index)
# ci.surface(m1)
# 
# # fit an interaction-only model
# m2=siRSM(y,U,V,interaction.only=TRUE)
# plot(m2)
# ## End(Not run)